Solid State MRI Techniques

Even though bone water and phosphorus are beyond the capability of conventional MRI techniques due to their extremely short transverse relaxation times, specially designed imaging sequences featuring extraordinarily short TE (solid-state imaging sequence) can be utilized. A couple of solid-state MRI techniques have been developed in response to challenges posed by species with extremely short effective transverse relaxation time.

2.2.1 Ultra-short Echo Time Imaging (UTE)

One such technique is ultra-short echo time (UTE) imaging [45]. In a typical UTE experiment (Figure 2.6a), the encoding gradient is switched on immediately after the transmit/receive dead time, and data acquisition is initiated on the ramp-up stage of the gradient. Each excitation and readout results in a half projection within the k-space. The orientation of the encoding gradient is altered from one TR to the next to cover the entire k-space (Figure 2.7a). As a result of ramp sampling, where the gradient is not at full strength, and k-space is traversed more slowly (but at an accelerating speed) as compared to when the gradient is fully ramped up, the center of k-space is more densely sampled as compared to the outer k-space.

Figure 2.6 Sequence diagrams of a) UTE and b) ZTE.

A UTE sequence is characterized by the following features: first, with effective echo time limited only by the hardware’s transmit/receive switch time (or receiver dead time), UTE is able to visualize spins with T2* as short as a few hundred microseconds [45, 46];

Figure 2.7 k-Space of a) UTE and b) ZTE-PETRA sequence. In UTE, k-space encoding starts during the ramp-up stage of the gradient. Therefore, sampling density is higher near the k- space center than in the outer region where the gradient reaches full strength. Encoding direction is altered from one TR to the next to cover the entire k-space. In ZTE, on the other hand, the gradient is fully ramped up prior to RF excitation, resulting in k-space data missing around the center due to the transmit/receive switch. In the case of ZTE-PETRA, the missing data points are recovered via single point imaging on a Cartesian grid through a second acquisition.

2.2.2 Zero Echo Time Imaging (ZTE)

Another powerful short-T2 visualization technique is zero echo time (ZTE) imaging

(Figure 2.6b). In contrast to UTE, the encoding gradient is fully ramped up prior to the RF pulse in ZTE, and data are collected immediately following the transmit/receive dead time. Since the gradient is always on, k-space encoding starts instantaneously after excitation (hence the name ‘zero’ echo time) and the data points encoded during dead time are lost. The missing data points are recovered via a second set of scans, where the inner k-space is filled either by FIDs acquired with lower bandwidth, as the SMRI [47]

used in WASPI [48], or with single point encoded with an incrementally-stepped gradient as in PETRA [49]. As shown in Figure 2.7b, a ZTE (PETRA in this case) k-space

consists of two distinct regions. The peripheral portion corresponds to the initial radial acquisition, while the central portion corresponds to the second acquisition acquired on a Cartesian grid.

Although the effective echo time of ZTE is also determined by the hardware, as in UTE, the center portion of k-space is traversed faster because the encoding gradient is already at full strength. Further, the PETRA acquisition allows the data points near k-space center to be acquired at the same effective echo time. Therefore, ZTE images possess higher signal-to-noise ratio (SNR) and are less prone to blurring than UTE images.

One inherent drawback of ZTE is the presence of encoding gradient during excitation, which renders the RF pulse spatially selective. The varying excitation profile from k- space line to k-space line can result in undesirable shading artifacts and blurring

throughout the reconstructed image. Since the frequency response of a RF pulse could be approximated by the Fourier transform of the pulse shape in the low flip-angle regime [50], the wider the pulse, the heavier the signal modulation. As well, restrictions on the specific absorption ratio (SAR) limit the peak B1 power of the RF pulse. As a result, ZTE

is restricted to low FA applications in humans. 2.2.3 UTE and ZTE Image Reconstruction

application of FFT will not work. Instead, the radially collected data points must be first re-gridded onto a Cartesian coordinate. The re-gridded points are essentially the weighted sum of the neighboring original data. However, the varied distance of each raw data point to a specific point on the Cartesian grid determines they may not necessarily have the same contribution in this process. For instance, in Figure 2.7, the sampling pattern is generally denser in the inner k-space than in the outer. Therefore, a faithful image reconstruction relies on proper weight (or sampling density) compensation of the raw data [51]. Following the re-gridding procedure, the image is generated with FFT as in conventional MRI.

2.3 Separation of Different Bone Water Pools